AI Became Tool for New Generation of Scammers: How Deepfakes Steal Millions of Dollars — Study

• TRM Labs recorded 263% higher losses from deepfake scams in the first months of 2026 compared with the total for all of last year.
• In particular, attackers are cloning voices, faces, and documents.
• Detection recommendations include specific visual, audio, and behavioral indicators.

In 2026, deepfakes have evolved from an experimental technology into a tool for fraud, theft, KYC bypass, and the creation of illegal content. According to TRM Labs, the AI-in-crime adoption index rose from 28 points in 2024 to 54 in 2026, and losses from deepfake scams in the first months of 2026 have already exceeded the total for all of 2025 by 263%. 

Against the backdrop of cheaper generation and the spread of accessible AI tools, the technology is simultaneously creating risks for the financial sector, crypto platforms, businesses, and everyday users.

How Deepfakes Are Used in Scams

One of the most high-profile cases occurred in January 2024 at the engineering firm Arup. During a video conference, an employee spoke with five “colleagues,” including the company’s UK-based chief financial officer. In reality, everyone on the call except the employee was a deepfake. Trusting the instructions he received, he approved 15 transfers totaling $25 million to accounts in Hong Kong.

According to Deloitte, losses from generative AI-enabled fraud in the US could rise from $12.3 billion in 2023 to $40 billion in 2027. In 2025–2026, the FBI also repeatedly warned about the use of synthetic video and voice to impersonate officials, company executives, and law enforcement officers.

Seconds Before the Transaction: How Crypto Fraud Is Changing in the Age of AI

21.04.2026

Read now

Key schemes include:

• Impersonating executives and relatives to obtain transfers, authentication codes, or confidential information
• Voice cloning in modern variations of the “grandchild scam”
• Romance scams, where AI generates photos and videos, and helps maintain many conversations at the same time
• Impersonation of law enforcement officers and government officials
• Creating synthetic identities and bypassing KYC checks at banks, fintech services, and cryptocurrency platforms
• Deepfake-as-a-Service, where the relevant tools are sold as a ready-made criminal service

TRM Labs identified providers offering standalone models for face swapping and voice cloning for about $500, while annual access to a deepfake toolset costs around $3,000. 

Hany Farid, a digital forensics researcher at the University of California, Berkeley, demonstrated that creating a real-time video impersonation may require only a single screenshot and $10 software.

The crypto industry poses a separate challenge: Farid’s research showed that video identity checks at one major cryptocurrency exchange, when conducted without human involvement, could be bypassed using synthetic video.

How to Spot a Deepfake

According to Farid, people identify deepfakes only slightly better than random guessing, and excessive confidence in one’s own assessment is often associated with lower accuracy. The FBI recommends first paying attention to simple visual, audio, and behavioral signs:

• Unnatural blinking, facial movement, or mouth movement
• Distorted hands, accessories, shadows, and lighting
• Voice and lip movement out of sync
• Unnatural intonation or speaking pace
• A demand to immediately transfer funds or provide confidential data
• A request to switch to an unusual communication channel

To verify someone’s identity, the FBI advises using an independent channel that is not connected to the original message. For family, you can agree in advance on a secret word, and in business — arrange a callback to a known number or involve a second employee to confirm financial transactions.

Top 15 neural networks for video creation: the best solutions from Google Veo 3.1 to Seedance

19.08.2026

Read now

The risks are already going beyond scams. In January 2026, Ashley St. Clair (the mother of one of Elon Musk’s children) filed a lawsuit against xAI, claiming that Grok created explicit deepfakes without her consent, including one depicting a minor. 

In June, British MP Jess Asato also filed a lawsuit against xAI over sexualized deepfakes created by Grok. The case could become one of the first UK proceedings to directly examine an AI developer’s liability for content generated by its model.

At the same time, companies are also working on protection technologies. In April, the World project, backed by OpenAI CEO Sam Altman, unveiled an update to World ID, positioning the system as infrastructure for “full proof of personhood” and protection against bots and deepfakes. Among potential use cases, the project named Tinder, Zoom, DocuSign, Razer, Okta, Vercel, and Browserbase.

Сообщение AI Became Tool for New Generation of Scammers: How Deepfakes Steal Millions of Dollars — Study появились сначала на INCRYPTED.